wimlib.net Git - wimlib/blob - src/win32_capture.c

   1 /*
   2  * win32_capture.c - Windows-specific code for capturing files into a WIM image.
   3  *
   4  * This now uses the native Windows NT API a lot and not just Win32.
   5  */
   6
   7 /*
   8  * Copyright (C) 2013, 2014, 2015 Eric Biggers
   9  *
  10  * This file is free software; you can redistribute it and/or modify it under
  11  * the terms of the GNU Lesser General Public License as published by the Free
  12  * Software Foundation; either version 3 of the License, or (at your option) any
  13  * later version.
  14  *
  15  * This file is distributed in the hope that it will be useful, but WITHOUT
  16  * ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or FITNESS
  17  * FOR A PARTICULAR PURPOSE. See the GNU Lesser General Public License for more
  18  * details.
  19  *
  20  * You should have received a copy of the GNU Lesser General Public License
  21  * along with this file; if not, see http://www.gnu.org/licenses/.
  22  */
  23
  24 #ifdef __WIN32__
  25
  26 #ifdef HAVE_CONFIG_H
  27 #  include "config.h"
  28 #endif
  29
  30 #include "wimlib/win32_common.h"
  31
  32 #include "wimlib/assert.h"
  33 #include "wimlib/blob_table.h"
  34 #include "wimlib/capture.h"
  35 #include "wimlib/dentry.h"
  36 #include "wimlib/encoding.h"
  37 #include "wimlib/endianness.h"
  38 #include "wimlib/error.h"
  39 #include "wimlib/paths.h"
  40 #include "wimlib/reparse.h"
  41 #include "wimlib/wof.h"
  42
  43 struct winnt_scan_ctx {
  44         u32 vol_flags;
  45         unsigned long num_get_sd_access_denied;
  46         unsigned long num_get_sacl_priv_notheld;
  47
  48         /* True if WOF is definitely not attached to the volume being scanned;
  49          * false if it may be  */
  50         bool wof_not_attached;
  51 };
  52
  53 static inline const wchar_t *
  54 printable_path(const wchar_t *full_path)
  55 {
  56         /* Skip over \\?\ or \??\  */
  57         return full_path + 4;
  58 }
  59
  60 /*
  61  * If cur_dir is not NULL, open an existing file relative to the already-open
  62  * directory cur_dir.
  63  *
  64  * Otherwise, open the file specified by @path, which must be a Windows NT
  65  * namespace path.
  66  */
  67 static NTSTATUS
  68 winnt_openat(HANDLE cur_dir, const wchar_t *path, size_t path_nchars,
  69              ACCESS_MASK perms, HANDLE *h_ret)
  70 {
  71         UNICODE_STRING name;
  72         OBJECT_ATTRIBUTES attr;
  73         IO_STATUS_BLOCK iosb;
  74         NTSTATUS status;
  75
  76         name.Length = path_nchars * sizeof(wchar_t);
  77         name.MaximumLength = name.Length + sizeof(wchar_t);
  78         name.Buffer = (wchar_t *)path;
  79
  80         attr.Length = sizeof(attr);
  81         attr.RootDirectory = cur_dir;
  82         attr.ObjectName = &name;
  83         attr.Attributes = 0;
  84         attr.SecurityDescriptor = NULL;
  85         attr.SecurityQualityOfService = NULL;
  86
  87 retry:
  88         status = (*func_NtOpenFile)(h_ret, perms, &attr, &iosb,
  89                                     FILE_SHARE_VALID_FLAGS,
  90                                     FILE_OPEN_REPARSE_POINT |
  91                                             FILE_OPEN_FOR_BACKUP_INTENT |
  92                                             FILE_SYNCHRONOUS_IO_NONALERT |
  93                                             FILE_SEQUENTIAL_ONLY);
  94         if (!NT_SUCCESS(status)) {
  95                 /* Try requesting fewer permissions  */
  96                 if (status == STATUS_ACCESS_DENIED ||
  97                     status == STATUS_PRIVILEGE_NOT_HELD) {
  98                         if (perms & ACCESS_SYSTEM_SECURITY) {
  99                                 perms &= ~ACCESS_SYSTEM_SECURITY;
 100                                 goto retry;
 101                         }
 102                         if (perms & READ_CONTROL) {
 103                                 perms &= ~READ_CONTROL;
 104                                 goto retry;
 105                         }
 106                 }
 107         }
 108         return status;
 109 }
 110
 111 /* Read the first @size bytes from the file, or named data stream of a file,
 112  * described by @blob.  */
 113 int
 114 read_winnt_stream_prefix(const struct blob_descriptor *blob, u64 size,
 115                          const struct read_blob_callbacks *cbs)
 116 {
 117         const wchar_t *path;
 118         HANDLE h;
 119         NTSTATUS status;
 120         u8 buf[BUFFER_SIZE];
 121         u64 bytes_remaining;
 122         int ret;
 123
 124         /* This is an NT namespace path.  */
 125         path = blob->file_on_disk;
 126
 127         status = winnt_openat(NULL, path, wcslen(path),
 128                               FILE_READ_DATA | SYNCHRONIZE, &h);
 129         if (!NT_SUCCESS(status)) {
 130                 winnt_error(status, L"\"%ls\": Can't open for reading",
 131                             printable_path(path));
 132                 return WIMLIB_ERR_OPEN;
 133         }
 134
 135         ret = 0;
 136         bytes_remaining = size;
 137         while (bytes_remaining) {
 138                 IO_STATUS_BLOCK iosb;
 139                 ULONG count;
 140                 ULONG bytes_read;
 141
 142                 count = min(sizeof(buf), bytes_remaining);
 143
 144                 status = (*func_NtReadFile)(h, NULL, NULL, NULL,
 145                                             &iosb, buf, count, NULL, NULL);
 146                 if (unlikely(!NT_SUCCESS(status))) {
 147                         if (status == STATUS_END_OF_FILE) {
 148                                 ERROR("\"%ls\": File was concurrently truncated",
 149                                       printable_path(path));
 150                                 ret = WIMLIB_ERR_CONCURRENT_MODIFICATION_DETECTED;
 151                         } else {
 152                                 winnt_error(status, L"\"%ls\": Error reading data",
 153                                             printable_path(path));
 154                                 ret = WIMLIB_ERR_READ;
 155                         }
 156                         break;
 157                 }
 158
 159                 bytes_read = iosb.Information;
 160
 161                 bytes_remaining -= bytes_read;
 162                 ret = call_consume_chunk(buf, bytes_read, cbs);
 163                 if (ret)
 164                         break;
 165         }
 166         (*func_NtClose)(h);
 167         return ret;
 168 }
 169
 170 struct win32_encrypted_read_ctx {
 171         const struct read_blob_callbacks *cbs;
 172         int wimlib_err_code;
 173         u64 bytes_remaining;
 174 };
 175
 176 static DWORD WINAPI
 177 win32_encrypted_export_cb(unsigned char *data, void *_ctx, unsigned long len)
 178 {
 179         struct win32_encrypted_read_ctx *ctx = _ctx;
 180         int ret;
 181         size_t bytes_to_consume = min(len, ctx->bytes_remaining);
 182
 183         if (bytes_to_consume == 0)
 184                 return ERROR_SUCCESS;
 185
 186         ret = call_consume_chunk(data, bytes_to_consume, ctx->cbs);
 187         if (ret) {
 188                 ctx->wimlib_err_code = ret;
 189                 /* It doesn't matter what error code is returned here, as long
 190                  * as it isn't ERROR_SUCCESS.  */
 191                 return ERROR_READ_FAULT;
 192         }
 193         ctx->bytes_remaining -= bytes_to_consume;
 194         return ERROR_SUCCESS;
 195 }
 196
 197 int
 198 read_win32_encrypted_file_prefix(const struct blob_descriptor *blob,
 199                                  u64 size,
 200                                  const struct read_blob_callbacks *cbs)
 201 {
 202         struct win32_encrypted_read_ctx export_ctx;
 203         DWORD err;
 204         void *file_ctx;
 205         int ret;
 206         DWORD flags = 0;
 207
 208         if (blob->file_inode->i_attributes & FILE_ATTRIBUTE_DIRECTORY)
 209                 flags |= CREATE_FOR_DIR;
 210
 211         export_ctx.cbs = cbs;
 212         export_ctx.wimlib_err_code = 0;
 213         export_ctx.bytes_remaining = size;
 214
 215         err = OpenEncryptedFileRaw(blob->file_on_disk, flags, &file_ctx);
 216         if (err != ERROR_SUCCESS) {
 217                 win32_error(err,
 218                             L"Failed to open encrypted file \"%ls\" for raw read",
 219                             printable_path(blob->file_on_disk));
 220                 return WIMLIB_ERR_OPEN;
 221         }
 222         err = ReadEncryptedFileRaw(win32_encrypted_export_cb,
 223                                    &export_ctx, file_ctx);
 224         if (err != ERROR_SUCCESS) {
 225                 ret = export_ctx.wimlib_err_code;
 226                 if (ret == 0) {
 227                         win32_error(err,
 228                                     L"Failed to read encrypted file \"%ls\"",
 229                                     printable_path(blob->file_on_disk));
 230                         ret = WIMLIB_ERR_READ;
 231                 }
 232         } else if (export_ctx.bytes_remaining != 0) {
 233                 ERROR("Only could read %"PRIu64" of %"PRIu64" bytes from "
 234                       "encrypted file \"%ls\"",
 235                       size - export_ctx.bytes_remaining, size,
 236                       printable_path(blob->file_on_disk));
 237                 ret = WIMLIB_ERR_READ;
 238         } else {
 239                 ret = 0;
 240         }
 241         CloseEncryptedFileRaw(file_ctx);
 242         return ret;
 243 }
 244
 245 /*
 246  * Load the short name of a file into a WIM dentry.
 247  */
 248 static noinline_for_stack NTSTATUS
 249 winnt_get_short_name(HANDLE h, struct wim_dentry *dentry)
 250 {
 251         /* It's not any harder to just make the NtQueryInformationFile() system
 252          * call ourselves, and it saves a dumb call to FindFirstFile() which of
 253          * course has to create its own handle.  */
 254         NTSTATUS status;
 255         IO_STATUS_BLOCK iosb;
 256         u8 buf[128] _aligned_attribute(8);
 257         const FILE_NAME_INFORMATION *info;
 258
 259         status = (*func_NtQueryInformationFile)(h, &iosb, buf, sizeof(buf),
 260                                                 FileAlternateNameInformation);
 261         info = (const FILE_NAME_INFORMATION *)buf;
 262         if (NT_SUCCESS(status) && info->FileNameLength != 0) {
 263                 dentry->d_short_name = utf16le_dupz(info->FileName,
 264                                                     info->FileNameLength);
 265                 if (!dentry->d_short_name)
 266                         return STATUS_NO_MEMORY;
 267                 dentry->d_short_name_nbytes = info->FileNameLength;
 268         }
 269         return status;
 270 }
 271
 272 /*
 273  * Load the security descriptor of a file into the corresponding inode and the
 274  * WIM image's security descriptor set.
 275  */
 276 static noinline_for_stack NTSTATUS
 277 winnt_get_security_descriptor(HANDLE h, struct wim_inode *inode,
 278                               struct wim_sd_set *sd_set,
 279                               struct winnt_scan_ctx *ctx, int add_flags)
 280 {
 281         SECURITY_INFORMATION requestedInformation;
 282         u8 _buf[4096] _aligned_attribute(8);
 283         u8 *buf;
 284         ULONG bufsize;
 285         ULONG len_needed;
 286         NTSTATUS status;
 287
 288         /*
 289          * LABEL_SECURITY_INFORMATION is needed on Windows Vista and 7 because
 290          * Microsoft decided to add mandatory integrity labels to the SACL but
 291          * not have them returned by SACL_SECURITY_INFORMATION.
 292          *
 293          * BACKUP_SECURITY_INFORMATION is needed on Windows 8 because Microsoft
 294          * decided to add even more stuff to the SACL and still not have it
 295          * returned by SACL_SECURITY_INFORMATION; but they did remember that
 296          * backup applications exist and simply want to read the stupid thing
 297          * once and for all, so they added a flag to read the entire security
 298          * descriptor.
 299          *
 300          * Older versions of Windows tolerate these new flags being passed in.
 301          */
 302         requestedInformation = OWNER_SECURITY_INFORMATION |
 303                                GROUP_SECURITY_INFORMATION |
 304                                DACL_SECURITY_INFORMATION |
 305                                SACL_SECURITY_INFORMATION |
 306                                LABEL_SECURITY_INFORMATION |
 307                                BACKUP_SECURITY_INFORMATION;
 308
 309         buf = _buf;
 310         bufsize = sizeof(_buf);
 311
 312         /*
 313          * We need the file's security descriptor in
 314          * SECURITY_DESCRIPTOR_RELATIVE format, and we currently have a handle
 315          * opened with as many relevant permissions as possible.  At this point,
 316          * on Windows there are a number of options for reading a file's
 317          * security descriptor:
 318          *
 319          * GetFileSecurity():  This takes in a path and returns the
 320          * SECURITY_DESCRIPTOR_RELATIVE.  Problem: this uses an internal handle,
 321          * not ours, and the handle created internally doesn't specify
 322          * FILE_FLAG_BACKUP_SEMANTICS.  Therefore there can be access denied
 323          * errors on some files and directories, even when running as the
 324          * Administrator.
 325          *
 326          * GetSecurityInfo():  This takes in a handle and returns the security
 327          * descriptor split into a bunch of different parts.  This should work,
 328          * but it's dumb because we have to put the security descriptor back
 329          * together again.
 330          *
 331          * BackupRead():  This can read the security descriptor, but this is a
 332          * difficult-to-use API, probably only works as the Administrator, and
 333          * the format of the returned data is not well documented.
 334          *
 335          * NtQuerySecurityObject():  This is exactly what we need, as it takes
 336          * in a handle and returns the security descriptor in
 337          * SECURITY_DESCRIPTOR_RELATIVE format.  Only problem is that it's a
 338          * ntdll function and therefore not officially part of the Win32 API.
 339          * Oh well.
 340          */
 341         while (!(NT_SUCCESS(status = (*func_NtQuerySecurityObject)(h,
 342                                                                    requestedInformation,
 343                                                                    (PSECURITY_DESCRIPTOR)buf,
 344                                                                    bufsize,
 345                                                                    &len_needed))))
 346         {
 347                 switch (status) {
 348                 case STATUS_BUFFER_TOO_SMALL:
 349                         wimlib_assert(buf == _buf);
 350                         buf = MALLOC(len_needed);
 351                         if (!buf)
 352                                 return STATUS_NO_MEMORY;
 353                         bufsize = len_needed;
 354                         break;
 355                 case STATUS_PRIVILEGE_NOT_HELD:
 356                 case STATUS_ACCESS_DENIED:
 357                         if (add_flags & WIMLIB_ADD_FLAG_STRICT_ACLS) {
 358                 default:
 359                                 /* Permission denied in STRICT_ACLS mode, or
 360                                  * unknown error.  */
 361                                 goto out_free_buf;
 362                         }
 363                         if (requestedInformation & SACL_SECURITY_INFORMATION) {
 364                                 /* Try again without the SACL.  */
 365                                 ctx->num_get_sacl_priv_notheld++;
 366                                 requestedInformation &= ~(SACL_SECURITY_INFORMATION |
 367                                                           LABEL_SECURITY_INFORMATION |
 368                                                           BACKUP_SECURITY_INFORMATION);
 369                                 break;
 370                         }
 371                         /* Fake success (useful when capturing as
 372                          * non-Administrator).  */
 373                         ctx->num_get_sd_access_denied++;
 374                         status = STATUS_SUCCESS;
 375                         goto out_free_buf;
 376                 }
 377         }
 378
 379         /* Add the security descriptor to the WIM image, and save its ID in
 380          * file's inode.  */
 381         inode->i_security_id = sd_set_add_sd(sd_set, buf, len_needed);
 382         if (unlikely(inode->i_security_id < 0))
 383                 status = STATUS_NO_MEMORY;
 384 out_free_buf:
 385         if (unlikely(buf != _buf))
 386                 FREE(buf);
 387         return status;
 388 }
 389
 390 static int
 391 winnt_build_dentry_tree_recursive(struct wim_dentry **root_ret,
 392                                   HANDLE cur_dir,
 393                                   wchar_t *full_path,
 394                                   size_t full_path_nchars,
 395                                   const wchar_t *filename,
 396                                   size_t filename_nchars,
 397                                   struct capture_params *params,
 398                                   struct winnt_scan_ctx *ctx);
 399
 400 static int
 401 winnt_recurse_directory(HANDLE h,
 402                         wchar_t *full_path,
 403                         size_t full_path_nchars,
 404                         struct wim_dentry *parent,
 405                         struct capture_params *params,
 406                         struct winnt_scan_ctx *ctx)
 407 {
 408         void *buf;
 409         const size_t bufsize = 8192;
 410         IO_STATUS_BLOCK iosb;
 411         NTSTATUS status;
 412         int ret;
 413
 414         buf = MALLOC(bufsize);
 415         if (!buf)
 416                 return WIMLIB_ERR_NOMEM;
 417
 418         /* Using NtQueryDirectoryFile() we can re-use the same open handle,
 419          * which we opened with FILE_FLAG_BACKUP_SEMANTICS.  */
 420
 421         while (NT_SUCCESS(status = (*func_NtQueryDirectoryFile)(h, NULL, NULL, NULL,
 422                                                                 &iosb, buf, bufsize,
 423                                                                 FileNamesInformation,
 424                                                                 FALSE, NULL, FALSE)))
 425         {
 426                 const FILE_NAMES_INFORMATION *info = buf;
 427                 for (;;) {
 428                         if (!should_ignore_filename(info->FileName,
 429                                                     info->FileNameLength / 2))
 430                         {
 431                                 wchar_t *p;
 432                                 wchar_t *filename;
 433                                 struct wim_dentry *child;
 434
 435                                 p = full_path + full_path_nchars;
 436                                 /* Only add a backslash if we don't already have
 437                                  * one.  This prevents a duplicate backslash
 438                                  * from being added when the path to the capture
 439                                  * dir had a trailing backslash.  */
 440                                 if (*(p - 1) != L'\\')
 441                                         *p++ = L'\\';
 442                                 filename = p;
 443                                 p = wmempcpy(filename, info->FileName,
 444                                              info->FileNameLength / 2);
 445                                 *p = '\0';
 446
 447                                 ret = winnt_build_dentry_tree_recursive(
 448                                                         &child,
 449                                                         h,
 450                                                         full_path,
 451                                                         p - full_path,
 452                                                         filename,
 453                                                         info->FileNameLength / 2,
 454                                                         params,
 455                                                         ctx);
 456
 457                                 full_path[full_path_nchars] = L'\0';
 458
 459                                 if (ret)
 460                                         goto out_free_buf;
 461                                 attach_scanned_tree(parent, child, params->blob_table);
 462                         }
 463                         if (info->NextEntryOffset == 0)
 464                                 break;
 465                         info = (const FILE_NAMES_INFORMATION *)
 466                                         ((const u8 *)info + info->NextEntryOffset);
 467                 }
 468         }
 469
 470         if (unlikely(status != STATUS_NO_MORE_FILES)) {
 471                 winnt_error(status, L"\"%ls\": Can't read directory",
 472                             printable_path(full_path));
 473                 ret = WIMLIB_ERR_READ;
 474         }
 475 out_free_buf:
 476         FREE(buf);
 477         return ret;
 478 }
 479
 480 /* Reparse point fixup status code  */
 481 #define RP_FIXED        (-1)
 482
 483 static bool
 484 file_has_ino_and_dev(HANDLE h, u64 ino, u64 dev)
 485 {
 486         NTSTATUS status;
 487         IO_STATUS_BLOCK iosb;
 488         FILE_INTERNAL_INFORMATION int_info;
 489         FILE_FS_VOLUME_INFORMATION vol_info;
 490
 491         status = (*func_NtQueryInformationFile)(h, &iosb,
 492                                                 &int_info, sizeof(int_info),
 493                                                 FileInternalInformation);
 494         if (!NT_SUCCESS(status))
 495                 return false;
 496
 497         if (int_info.IndexNumber.QuadPart != ino)
 498                 return false;
 499
 500         status = (*func_NtQueryVolumeInformationFile)(h, &iosb,
 501                                                       &vol_info, sizeof(vol_info),
 502                                                       FileFsVolumeInformation);
 503         if (!(NT_SUCCESS(status) || status == STATUS_BUFFER_OVERFLOW))
 504                 return false;
 505
 506         if (iosb.Information <
 507              offsetof(FILE_FS_VOLUME_INFORMATION, VolumeSerialNumber) +
 508              sizeof(vol_info.VolumeSerialNumber))
 509                 return false;
 510
 511         return (vol_info.VolumeSerialNumber == dev);
 512 }
 513
 514 /*
 515  * This is the Windows equivalent of unix_relativize_link_target(); see there
 516  * for general details.  This version works with an "absolute" Windows link
 517  * target, specified from the root of the Windows kernel object namespace.  Note
 518  * that we have to open directories with a trailing slash when present because
 519  * \??\E: opens the E: device itself and not the filesystem root directory.
 520  */
 521 static const wchar_t *
 522 winnt_relativize_link_target(const wchar_t *target, size_t target_nbytes,
 523                              u64 ino, u64 dev)
 524 {
 525         UNICODE_STRING name;
 526         OBJECT_ATTRIBUTES attr;
 527         IO_STATUS_BLOCK iosb;
 528         NTSTATUS status;
 529         const wchar_t *target_end;
 530         const wchar_t *p;
 531
 532         target_end = target + (target_nbytes / sizeof(wchar_t));
 533
 534         /* Empty path??? */
 535         if (target_end == target)
 536                 return target;
 537
 538         /* No leading slash???  */
 539         if (target[0] != L'\\')
 540                 return target;
 541
 542         /* UNC path???  */
 543         if ((target_end - target) >= 2 &&
 544             target[0] == L'\\' && target[1] == L'\\')
 545                 return target;
 546
 547         attr.Length = sizeof(attr);
 548         attr.RootDirectory = NULL;
 549         attr.ObjectName = &name;
 550         attr.Attributes = 0;
 551         attr.SecurityDescriptor = NULL;
 552         attr.SecurityQualityOfService = NULL;
 553
 554         name.Buffer = (wchar_t *)target;
 555         name.Length = 0;
 556         p = target;
 557         do {
 558                 HANDLE h;
 559                 const wchar_t *orig_p = p;
 560
 561                 /* Skip non-backslashes  */
 562                 while (p != target_end && *p != L'\\')
 563                         p++;
 564
 565                 /* Skip backslashes  */
 566                 while (p != target_end && *p == L'\\')
 567                         p++;
 568
 569                 /* Append path component  */
 570                 name.Length += (p - orig_p) * sizeof(wchar_t);
 571                 name.MaximumLength = name.Length;
 572
 573                 /* Try opening the file  */
 574                 status = (*func_NtOpenFile) (&h,
 575                                              FILE_READ_ATTRIBUTES | FILE_TRAVERSE,
 576                                              &attr,
 577                                              &iosb,
 578                                              FILE_SHARE_VALID_FLAGS,
 579                                              FILE_OPEN_FOR_BACKUP_INTENT);
 580
 581                 if (NT_SUCCESS(status)) {
 582                         /* Reset root directory  */
 583                         if (attr.RootDirectory)
 584                                 (*func_NtClose)(attr.RootDirectory);
 585                         attr.RootDirectory = h;
 586                         name.Buffer = (wchar_t *)p;
 587                         name.Length = 0;
 588
 589                         if (file_has_ino_and_dev(h, ino, dev))
 590                                 goto out_close_root_dir;
 591                 }
 592         } while (p != target_end);
 593
 594         p = target;
 595
 596 out_close_root_dir:
 597         if (attr.RootDirectory)
 598                 (*func_NtClose)(attr.RootDirectory);
 599         while (p > target && *(p - 1) == L'\\')
 600                 p--;
 601         return p;
 602 }
 603
 604 static int
 605 winnt_rpfix_progress(struct capture_params *params, const wchar_t *path,
 606                      const struct link_reparse_point *link, int scan_status)
 607 {
 608         size_t print_name_nchars = link->print_name_nbytes / sizeof(wchar_t);
 609         wchar_t print_name0[print_name_nchars + 1];
 610
 611         wmemcpy(print_name0, link->print_name, print_name_nchars);
 612         print_name0[print_name_nchars] = L'\0';
 613
 614         params->progress.scan.cur_path = printable_path(path);
 615         params->progress.scan.symlink_target = print_name0;
 616         return do_capture_progress(params, scan_status, NULL);
 617 }
 618
 619 static int
 620 winnt_try_rpfix(struct reparse_buffer_disk *rpbuf, u16 *rpbuflen_p,
 621                 const wchar_t *path, struct capture_params *params)
 622 {
 623         struct link_reparse_point link;
 624         const wchar_t *rel_target;
 625         int ret;
 626
 627         if (parse_link_reparse_point(rpbuf, *rpbuflen_p, &link)) {
 628                 /* Couldn't understand the reparse data; don't do the fixup.  */
 629                 return 0;
 630         }
 631
 632         /*
 633          * Don't do reparse point fixups on relative symbolic links.
 634          *
 635          * On Windows, a relative symbolic link is supposed to be identifiable
 636          * by having reparse tag WIM_IO_REPARSE_TAG_SYMLINK and flags
 637          * SYMBOLIC_LINK_RELATIVE.  We will use this information, although this
 638          * may not always do what the user expects, since drive-relative
 639          * symbolic links such as "\Users\Public" have SYMBOLIC_LINK_RELATIVE
 640          * set, in addition to truly relative symbolic links such as "Users" or
 641          * "Users\Public".  However, WIMGAPI (as of Windows 8.1) has this same
 642          * behavior.
 643          *
 644          * Otherwise, as far as I can tell, the targets of symbolic links that
 645          * are NOT relative, as well as junctions (note: a mountpoint is the
 646          * sames thing as a junction), must be NT namespace paths, for example:
 647          *
 648          *     - \??\e:\Users\Public
 649          *     - \DosDevices\e:\Users\Public
 650          *     - \Device\HardDiskVolume4\Users\Public
 651          *     - \??\Volume{c47cb07c-946e-4155-b8f7-052e9cec7628}\Users\Public
 652          *     - \DosDevices\Volume{c47cb07c-946e-4155-b8f7-052e9cec7628}\Users\Public
 653          */
 654         if (link_is_relative_symlink(&link))
 655                 return 0;
 656
 657         rel_target = winnt_relativize_link_target(link.substitute_name,
 658                                                   link.substitute_name_nbytes,
 659                                                   params->capture_root_ino,
 660                                                   params->capture_root_dev);
 661
 662         if (rel_target == link.substitute_name) {
 663                 /* Target points outside of the tree being captured or had an
 664                  * unrecognized path format.  Don't adjust it.  */
 665                 return winnt_rpfix_progress(params, path, &link,
 666                                             WIMLIB_SCAN_DENTRY_NOT_FIXED_SYMLINK);
 667         }
 668
 669         /* We have an absolute target pointing within the directory being
 670          * captured. @rel_target is the suffix of the link target that is the
 671          * part relative to the directory being captured.
 672          *
 673          * We will cut off the prefix before this part (which is the path to the
 674          * directory being captured) and add a dummy prefix.  Since the process
 675          * will need to be reversed when applying the image, it doesn't matter
 676          * what exactly the prefix is, as long as it looks like an absolute
 677          * path.  */
 678
 679         static const wchar_t prefix[6] = L"\\??\\X:";
 680         static const size_t num_unprintable_chars = 4;
 681
 682         size_t rel_target_nbytes =
 683                 link.substitute_name_nbytes - ((const u8 *)rel_target -
 684                                                (const u8 *)link.substitute_name);
 685
 686         wchar_t tmp[(sizeof(prefix) + rel_target_nbytes) / sizeof(wchar_t)];
 687
 688         memcpy(tmp, prefix, sizeof(prefix));
 689         memcpy(tmp + ARRAY_LEN(prefix), rel_target, rel_target_nbytes);
 690
 691         link.substitute_name = tmp;
 692         link.substitute_name_nbytes = sizeof(tmp);
 693
 694         link.print_name = link.substitute_name + num_unprintable_chars;
 695         link.print_name_nbytes = link.substitute_name_nbytes -
 696                                  (num_unprintable_chars * sizeof(wchar_t));
 697
 698         if (make_link_reparse_point(&link, rpbuf, rpbuflen_p))
 699                 return 0;
 700
 701         ret = winnt_rpfix_progress(params, path, &link,
 702                                    WIMLIB_SCAN_DENTRY_FIXED_SYMLINK);
 703         if (ret)
 704                 return ret;
 705         return RP_FIXED;
 706 }
 707
 708 /* Load the reparse data of a file into the corresponding WIM inode.  If the
 709  * reparse point is a symbolic link or junction with an absolute target and
 710  * RPFIX mode is enabled, then also rewrite its target to be relative to the
 711  * capture root.  */
 712 static noinline_for_stack int
 713 winnt_load_reparse_data(HANDLE h, struct wim_inode *inode,
 714                         const wchar_t *full_path, struct capture_params *params)
 715 {
 716         struct reparse_buffer_disk rpbuf;
 717         DWORD bytes_returned;
 718         u16 rpbuflen;
 719         int ret;
 720
 721         if (inode->i_attributes & FILE_ATTRIBUTE_ENCRYPTED) {
 722                 /* See comment above assign_stream_types_encrypted()  */
 723                 WARNING("Ignoring reparse data of encrypted file \"%ls\"",
 724                         printable_path(full_path));
 725                 return 0;
 726         }
 727
 728         if (!DeviceIoControl(h, FSCTL_GET_REPARSE_POINT,
 729                              NULL, 0, &rpbuf, REPARSE_POINT_MAX_SIZE,
 730                              &bytes_returned, NULL))
 731         {
 732                 win32_error(GetLastError(), L"\"%ls\": Can't get reparse point",
 733                             printable_path(full_path));
 734                 return WIMLIB_ERR_READLINK;
 735         }
 736
 737         rpbuflen = bytes_returned;
 738
 739         if (unlikely(rpbuflen < REPARSE_DATA_OFFSET)) {
 740                 ERROR("\"%ls\": reparse point buffer is too short",
 741                       printable_path(full_path));
 742                 return WIMLIB_ERR_INVALID_REPARSE_DATA;
 743         }
 744
 745         if (params->add_flags & WIMLIB_ADD_FLAG_RPFIX) {
 746                 ret = winnt_try_rpfix(&rpbuf, &rpbuflen, full_path, params);
 747                 if (ret == RP_FIXED)
 748                         inode->i_rp_flags &= ~WIM_RP_FLAG_NOT_FIXED;
 749                 else if (ret)
 750                         return ret;
 751         }
 752
 753         inode->i_reparse_tag = le32_to_cpu(rpbuf.rptag);
 754         inode->i_rp_reserved = le16_to_cpu(rpbuf.rpreserved);
 755
 756         if (!inode_add_stream_with_data(inode,
 757                                         STREAM_TYPE_REPARSE_POINT,
 758                                         NO_STREAM_NAME,
 759                                         rpbuf.rpdata,
 760                                         rpbuflen - REPARSE_DATA_OFFSET,
 761                                         params->blob_table))
 762                 return WIMLIB_ERR_NOMEM;
 763
 764         return 0;
 765 }
 766
 767 static DWORD WINAPI
 768 win32_tally_encrypted_size_cb(unsigned char *_data, void *_size_ret,
 769                               unsigned long len)
 770 {
 771         *(u64*)_size_ret += len;
 772         return ERROR_SUCCESS;
 773 }
 774
 775 static int
 776 win32_get_encrypted_file_size(const wchar_t *path, bool is_dir, u64 *size_ret)
 777 {
 778         DWORD err;
 779         void *file_ctx;
 780         int ret;
 781         DWORD flags = 0;
 782
 783         if (is_dir)
 784                 flags |= CREATE_FOR_DIR;
 785
 786         err = OpenEncryptedFileRaw(path, flags, &file_ctx);
 787         if (err != ERROR_SUCCESS) {
 788                 win32_error(err,
 789                             L"Failed to open encrypted file \"%ls\" for raw read",
 790                             printable_path(path));
 791                 return WIMLIB_ERR_OPEN;
 792         }
 793         *size_ret = 0;
 794         err = ReadEncryptedFileRaw(win32_tally_encrypted_size_cb,
 795                                    size_ret, file_ctx);
 796         if (err != ERROR_SUCCESS) {
 797                 win32_error(err,
 798                             L"Failed to read raw encrypted data from \"%ls\"",
 799                             printable_path(path));
 800                 ret = WIMLIB_ERR_READ;
 801         } else {
 802                 ret = 0;
 803         }
 804         CloseEncryptedFileRaw(file_ctx);
 805         return ret;
 806 }
 807
 808 static int
 809 winnt_scan_efsrpc_raw_data(struct wim_inode *inode, const wchar_t *nt_path,
 810                            struct list_head *unhashed_blobs)
 811 {
 812         struct blob_descriptor *blob;
 813         struct wim_inode_stream *strm;
 814         int ret;
 815
 816         blob = new_blob_descriptor();
 817         if (!blob)
 818                 goto err_nomem;
 819
 820         blob->file_on_disk = WCSDUP(nt_path);
 821         if (!blob->file_on_disk)
 822                 goto err_nomem;
 823         blob->blob_location = BLOB_WIN32_ENCRYPTED;
 824
 825         /* OpenEncryptedFileRaw() expects a Win32 name.  */
 826         wimlib_assert(!wmemcmp(blob->file_on_disk, L"\\??\\", 4));
 827         blob->file_on_disk[1] = L'\\';
 828
 829         blob->file_inode = inode;
 830
 831         ret = win32_get_encrypted_file_size(blob->file_on_disk,
 832                                             (inode->i_attributes & FILE_ATTRIBUTE_DIRECTORY),
 833                                             &blob->size);
 834         if (ret)
 835                 goto err;
 836
 837         /* Empty EFSRPC data does not make sense  */
 838         wimlib_assert(blob->size != 0);
 839
 840         strm = inode_add_stream(inode, STREAM_TYPE_EFSRPC_RAW_DATA,
 841                                 NO_STREAM_NAME, blob);
 842         if (!strm)
 843                 goto err_nomem;
 844
 845         prepare_unhashed_blob(blob, inode, strm->stream_id, unhashed_blobs);
 846         return 0;
 847
 848 err_nomem:
 849         ret = WIMLIB_ERR_NOMEM;
 850 err:
 851         free_blob_descriptor(blob);
 852         return ret;
 853 }
 854
 855 static bool
 856 get_data_stream_name(wchar_t *raw_stream_name, size_t raw_stream_name_nchars,
 857                      wchar_t **stream_name_ret, size_t *stream_name_nchars_ret)
 858 {
 859         const wchar_t *sep, *type, *end;
 860
 861         /* The stream name should be returned as :NAME:TYPE  */
 862         if (raw_stream_name_nchars < 1)
 863                 return false;
 864         if (raw_stream_name[0] != L':')
 865                 return false;
 866
 867         raw_stream_name++;
 868         raw_stream_name_nchars--;
 869
 870         end = raw_stream_name + raw_stream_name_nchars;
 871
 872         sep = wmemchr(raw_stream_name, L':', raw_stream_name_nchars);
 873         if (!sep)
 874                 return false;
 875
 876         type = sep + 1;
 877         if (end - type != 5)
 878                 return false;
 879
 880         if (wmemcmp(type, L"$DATA", 5))
 881                 return false;
 882
 883         *stream_name_ret = raw_stream_name;
 884         *stream_name_nchars_ret = sep - raw_stream_name;
 885         return true;
 886 }
 887
 888 /* Build the path to the data stream.  For unnamed streams, this is simply the
 889  * path to the file.  For named streams, this is the path to the file, followed
 890  * by a colon, followed by the stream name.  */
 891 static wchar_t *
 892 build_data_stream_path(const wchar_t *path, size_t path_nchars,
 893                        const wchar_t *stream_name, size_t stream_name_nchars)
 894 {
 895         size_t stream_path_nchars;
 896         wchar_t *stream_path;
 897         wchar_t *p;
 898
 899         stream_path_nchars = path_nchars;
 900         if (stream_name_nchars)
 901                 stream_path_nchars += 1 + stream_name_nchars;
 902
 903         stream_path = MALLOC((stream_path_nchars + 1) * sizeof(wchar_t));
 904         if (stream_path) {
 905                 p = wmempcpy(stream_path, path, path_nchars);
 906                 if (stream_name_nchars) {
 907                         *p++ = L':';
 908                         p = wmempcpy(p, stream_name, stream_name_nchars);
 909                 }
 910                 *p++ = L'\0';
 911         }
 912         return stream_path;
 913 }
 914
 915 static int
 916 winnt_scan_data_stream(const wchar_t *path, size_t path_nchars,
 917                        wchar_t *raw_stream_name, size_t raw_stream_name_nchars,
 918                        u64 stream_size,
 919                        struct wim_inode *inode, struct list_head *unhashed_blobs)
 920 {
 921         wchar_t *stream_name;
 922         size_t stream_name_nchars;
 923         struct blob_descriptor *blob;
 924         struct wim_inode_stream *strm;
 925
 926         /* Given the raw stream name (which is something like
 927          * :streamname:$DATA), extract just the stream name part (streamname).
 928          * Ignore any non-$DATA streams.  */
 929         if (!get_data_stream_name(raw_stream_name, raw_stream_name_nchars,
 930                                   &stream_name, &stream_name_nchars))
 931                 return 0;
 932
 933         stream_name[stream_name_nchars] = L'\0';
 934
 935         /* If the stream is non-empty, set up a blob descriptor for it.  */
 936         if (stream_size != 0) {
 937                 blob = new_blob_descriptor();
 938                 if (!blob)
 939                         goto err_nomem;
 940                 blob->file_on_disk = build_data_stream_path(path,
 941                                                             path_nchars,
 942                                                             stream_name,
 943                                                             stream_name_nchars);
 944                 if (!blob->file_on_disk)
 945                         goto err_nomem;
 946                 blob->blob_location = BLOB_IN_WINNT_FILE_ON_DISK;
 947                 blob->size = stream_size;
 948                 blob->file_inode = inode;
 949         } else {
 950                 blob = NULL;
 951         }
 952
 953         strm = inode_add_stream(inode, STREAM_TYPE_DATA, stream_name, blob);
 954         if (!strm)
 955                 goto err_nomem;
 956
 957         prepare_unhashed_blob(blob, inode, strm->stream_id, unhashed_blobs);
 958         return 0;
 959
 960 err_nomem:
 961         free_blob_descriptor(blob);
 962         return WIMLIB_ERR_NOMEM;
 963 }
 964
 965 /*
 966  * Load information about the data streams of an open file into a WIM inode.
 967  *
 968  * We use the NtQueryInformationFile() system call instead of FindFirstStream()
 969  * and FindNextStream().  This is done for two reasons:
 970  *
 971  * - FindFirstStream() opens its own handle to the file or directory and
 972  *   apparently does so without specifying FILE_FLAG_BACKUP_SEMANTICS, thereby
 973  *   causing access denied errors on certain files (even when running as the
 974  *   Administrator).
 975  * - FindFirstStream() and FindNextStream() is only available on Windows Vista
 976  *   and later, whereas the stream support in NtQueryInformationFile() was
 977  *   already present in Windows XP.
 978  */
 979 static noinline_for_stack int
 980 winnt_scan_data_streams(HANDLE h, const wchar_t *path, size_t path_nchars,
 981                         struct wim_inode *inode, struct list_head *unhashed_blobs,
 982                         u64 file_size, u32 vol_flags)
 983 {
 984         int ret;
 985         u8 _buf[4096] _aligned_attribute(8);
 986         u8 *buf;
 987         size_t bufsize;
 988         IO_STATUS_BLOCK iosb;
 989         NTSTATUS status;
 990         FILE_STREAM_INFORMATION *info;
 991
 992         buf = _buf;
 993         bufsize = sizeof(_buf);
 994
 995         if (!(vol_flags & FILE_NAMED_STREAMS))
 996                 goto unnamed_only;
 997
 998         /* Get a buffer containing the stream information.  */
 999         while (!NT_SUCCESS(status = (*func_NtQueryInformationFile)(h,
1000                                                                    &iosb,
1001                                                                    buf,
1002                                                                    bufsize,
1003                                                                    FileStreamInformation)))
1004         {
1005
1006                 switch (status) {
1007                 case STATUS_BUFFER_OVERFLOW:
1008                         {
1009                                 u8 *newbuf;
1010
1011                                 bufsize *= 2;
1012                                 if (buf == _buf)
1013                                         newbuf = MALLOC(bufsize);
1014                                 else
1015                                         newbuf = REALLOC(buf, bufsize);
1016                                 if (!newbuf) {
1017                                         ret = WIMLIB_ERR_NOMEM;
1018                                         goto out_free_buf;
1019                                 }
1020                                 buf = newbuf;
1021                         }
1022                         break;
1023                 case STATUS_NOT_IMPLEMENTED:
1024                 case STATUS_NOT_SUPPORTED:
1025                 case STATUS_INVALID_INFO_CLASS:
1026                         goto unnamed_only;
1027                 default:
1028                         winnt_error(status,
1029                                     L"\"%ls\": Failed to query stream information",
1030                                     printable_path(path));
1031                         ret = WIMLIB_ERR_READ;
1032                         goto out_free_buf;
1033                 }
1034         }
1035
1036         if (iosb.Information == 0) {
1037                 /* No stream information.  */
1038                 ret = 0;
1039                 goto out_free_buf;
1040         }
1041
1042         /* Parse one or more stream information structures.  */
1043         info = (FILE_STREAM_INFORMATION *)buf;
1044         for (;;) {
1045                 /* Load the stream information.  */
1046                 ret = winnt_scan_data_stream(path, path_nchars,
1047                                              info->StreamName,
1048                                              info->StreamNameLength / 2,
1049                                              info->StreamSize.QuadPart,
1050                                              inode, unhashed_blobs);
1051                 if (ret)
1052                         goto out_free_buf;
1053
1054                 if (info->NextEntryOffset == 0) {
1055                         /* No more stream information.  */
1056                         break;
1057                 }
1058                 /* Advance to next stream information.  */
1059                 info = (FILE_STREAM_INFORMATION *)
1060                                 ((u8 *)info + info->NextEntryOffset);
1061         }
1062         ret = 0;
1063         goto out_free_buf;
1064
1065 unnamed_only:
1066         /* The volume does not support named streams.  Only capture the unnamed
1067          * data stream.  */
1068         if (inode->i_attributes & (FILE_ATTRIBUTE_DIRECTORY |
1069                                    FILE_ATTRIBUTE_REPARSE_POINT))
1070         {
1071                 ret = 0;
1072                 goto out_free_buf;
1073         }
1074
1075         {
1076                 wchar_t stream_name[] = L"::$DATA";
1077                 ret = winnt_scan_data_stream(path, path_nchars, stream_name, 7,
1078                                              file_size, inode, unhashed_blobs);
1079         }
1080 out_free_buf:
1081         /* Free buffer if allocated on heap.  */
1082         if (unlikely(buf != _buf))
1083                 FREE(buf);
1084         return ret;
1085 }
1086
1087 static noinline_for_stack u64
1088 get_sort_key(HANDLE h)
1089 {
1090         STARTING_VCN_INPUT_BUFFER in = { .StartingVcn.QuadPart = 0 };
1091         RETRIEVAL_POINTERS_BUFFER out;
1092         DWORD bytesReturned;
1093
1094         if (!DeviceIoControl(h, FSCTL_GET_RETRIEVAL_POINTERS,
1095                              &in, sizeof(in),
1096                              &out, sizeof(out),
1097                              &bytesReturned, NULL))
1098                 return 0;
1099
1100         if (out.ExtentCount < 1)
1101                 return 0;
1102
1103         return out.Extents[0].Lcn.QuadPart;
1104 }
1105
1106 static void
1107 set_sort_key(struct wim_inode *inode, u64 sort_key)
1108 {
1109         for (unsigned i = 0; i < inode->i_num_streams; i++) {
1110                 struct wim_inode_stream *strm = &inode->i_streams[i];
1111                 struct blob_descriptor *blob = stream_blob_resolved(strm);
1112                 if (blob && (blob->blob_location == BLOB_IN_WINNT_FILE_ON_DISK ||
1113                              blob->blob_location == BLOB_WIN32_ENCRYPTED))
1114                         blob->sort_key = sort_key;
1115         }
1116 }
1117
1118 static inline bool
1119 should_try_to_use_wimboot_hash(const struct wim_inode *inode,
1120                                const struct winnt_scan_ctx *ctx,
1121                                const struct capture_params *params)
1122 {
1123         /* Directories and encrypted files aren't valid for external backing. */
1124         if (inode->i_attributes & (FILE_ATTRIBUTE_DIRECTORY |
1125                                    FILE_ATTRIBUTE_ENCRYPTED))
1126                 return false;
1127
1128         /* If the file is a reparse point, then try the hash fixup if it's a WOF
1129          * reparse point and we're in WIMBOOT mode.  Otherwise, try the hash
1130          * fixup if WOF may be attached. */
1131         if (inode->i_attributes & FILE_ATTRIBUTE_REPARSE_POINT)
1132                 return (inode->i_reparse_tag == WIM_IO_REPARSE_TAG_WOF) &&
1133                         (params->add_flags & WIMLIB_ADD_FLAG_WIMBOOT);
1134         return !ctx->wof_not_attached;
1135 }
1136
1137 /*
1138  * This function implements an optimization for capturing files from a
1139  * filesystem with a backing WIM(s).  If a file is WIM-backed, then we can
1140  * retrieve the SHA-1 message digest of its original contents from its reparse
1141  * point.  This may eliminate the need to read the file's data and/or allow the
1142  * file's data to be immediately deduplicated with existing data in the WIM.
1143  *
1144  * If WOF is attached, then this function is merely an optimization, but
1145  * potentially a very effective one.  If WOF is detached, then this function
1146  * really causes WIM-backed files to be, effectively, automatically
1147  * "dereferenced" when possible; the unnamed data stream is updated to reference
1148  * the original contents and the reparse point is removed.
1149  *
1150  * This function returns 0 if the fixup succeeded or was intentionally not
1151  * executed.  Otherwise it returns an error code.
1152  */
1153 static noinline_for_stack int
1154 try_to_use_wimboot_hash(HANDLE h, struct wim_inode *inode,
1155                         struct blob_table *blob_table,
1156                         struct winnt_scan_ctx *ctx, const wchar_t *full_path)
1157 {
1158         struct wim_inode_stream *reparse_strm = NULL;
1159         struct wim_inode_stream *strm;
1160         struct blob_descriptor *blob;
1161         u8 hash[SHA1_HASH_SIZE];
1162         int ret;
1163
1164         if (inode->i_attributes & FILE_ATTRIBUTE_REPARSE_POINT) {
1165                 struct reparse_buffer_disk rpbuf;
1166                 struct {
1167                         struct wof_external_info wof_info;
1168                         struct wim_provider_rpdata wim_info;
1169                 } *rpdata = (void *)rpbuf.rpdata;
1170                 struct blob_descriptor *reparse_blob;
1171
1172                 /* The file has a WOF reparse point, so WOF must be detached.
1173                  * We can read the reparse point directly.  */
1174                 ctx->wof_not_attached = true;
1175                 reparse_strm = inode_get_unnamed_stream(inode, STREAM_TYPE_REPARSE_POINT);
1176                 reparse_blob = stream_blob_resolved(reparse_strm);
1177
1178                 if (!reparse_blob || reparse_blob->size < sizeof(*rpdata))
1179                         return 0;  /* Not a WIM-backed file  */
1180
1181                 ret = read_blob_into_buf(reparse_blob, rpdata);
1182                 if (ret)
1183                         return ret;
1184
1185                 if (rpdata->wof_info.version != WOF_CURRENT_VERSION ||
1186                     rpdata->wof_info.provider != WOF_PROVIDER_WIM ||
1187                     rpdata->wim_info.version != 2)
1188                         return 0;  /* Not a WIM-backed file  */
1189
1190                 /* Okay, this is a WIM backed file.  Get its SHA-1 hash.  */
1191                 copy_hash(hash, rpdata->wim_info.unnamed_data_stream_hash);
1192         } else {
1193                 struct {
1194                         struct wof_external_info wof_info;
1195                         struct wim_provider_external_info wim_info;
1196                 } out;
1197                 IO_STATUS_BLOCK iosb;
1198                 NTSTATUS status;
1199
1200                 /* WOF may be attached.  Try reading this file's external
1201                  * backing info.  */
1202                 status = (*func_NtFsControlFile)(h, NULL, NULL, NULL, &iosb,
1203                                                  FSCTL_GET_EXTERNAL_BACKING,
1204                                                  NULL, 0, &out, sizeof(out));
1205
1206                 /* Is WOF not attached?  */
1207                 if (status == STATUS_INVALID_DEVICE_REQUEST) {
1208                         ctx->wof_not_attached = true;
1209                         return 0;
1210                 }
1211
1212                 /* Is this file not externally backed?  */
1213                 if (status == STATUS_OBJECT_NOT_EXTERNALLY_BACKED)
1214                         return 0;
1215
1216                 /* Does this file have an unknown type of external backing that
1217                  * needed a larger information buffer?  */
1218                 if (status == STATUS_BUFFER_TOO_SMALL)
1219                         return 0;
1220
1221                 /* Was there some other failure?  */
1222                 if (status != STATUS_SUCCESS) {
1223                         winnt_error(status,
1224                                     L"\"%ls\": FSCTL_GET_EXTERNAL_BACKING failed",
1225                                     full_path);
1226                         return WIMLIB_ERR_STAT;
1227                 }
1228
1229                 /* Is this file backed by a WIM?  */
1230                 if (out.wof_info.version != WOF_CURRENT_VERSION ||
1231                     out.wof_info.provider != WOF_PROVIDER_WIM ||
1232                     out.wim_info.version != WIM_PROVIDER_CURRENT_VERSION)
1233                         return 0;
1234
1235                 /* Okay, this is a WIM backed file.  Get its SHA-1 hash.  */
1236                 copy_hash(hash, out.wim_info.unnamed_data_stream_hash);
1237         }
1238
1239         /* If the file's unnamed data stream is nonempty, then fill in its hash
1240          * and deduplicate it if possible.
1241          *
1242          * With WOF detached, we require that the blob *must* de-duplicable for
1243          * any action can be taken, since without WOF we can't fall back to
1244          * getting the "dereferenced" data by reading the stream (the real
1245          * stream is sparse and contains all zeroes).  */
1246         strm = inode_get_unnamed_data_stream(inode);
1247         if (strm && (blob = stream_blob_resolved(strm))) {
1248                 struct blob_descriptor **back_ptr;
1249
1250                 if (reparse_strm && !lookup_blob(blob_table, hash))
1251                         return 0;
1252                 back_ptr = retrieve_pointer_to_unhashed_blob(blob);
1253                 copy_hash(blob->hash, hash);
1254                 if (after_blob_hashed(blob, back_ptr, blob_table) != blob)
1255                         free_blob_descriptor(blob);
1256         }
1257
1258         /* Remove the reparse point, if present.  */
1259         if (reparse_strm) {
1260                 inode_remove_stream(inode, reparse_strm, blob_table);
1261                 inode->i_attributes &= ~(FILE_ATTRIBUTE_REPARSE_POINT |
1262                                          FILE_ATTRIBUTE_SPARSE_FILE);
1263                 if (inode->i_attributes == 0)
1264                         inode->i_attributes = FILE_ATTRIBUTE_NORMAL;
1265         }
1266
1267         return 0;
1268 }
1269
1270 static noinline_for_stack u32
1271 get_volume_information(HANDLE h, const wchar_t *full_path,
1272                        struct capture_params *params)
1273 {
1274         FILE_FS_ATTRIBUTE_INFORMATION attr_info;
1275         FILE_FS_VOLUME_INFORMATION vol_info;
1276         IO_STATUS_BLOCK iosb;
1277         NTSTATUS status;
1278         u32 vol_flags;
1279
1280         /* Get volume flags  */
1281         status = (*func_NtQueryVolumeInformationFile)(h, &iosb,
1282                                                       &attr_info,
1283                                                       sizeof(attr_info),
1284                                                       FileFsAttributeInformation);
1285         if (likely((NT_SUCCESS(status) || status == STATUS_BUFFER_OVERFLOW) &&
1286                    (iosb.Information >=
1287                         offsetof(FILE_FS_ATTRIBUTE_INFORMATION,
1288                                  FileSystemAttributes) +
1289                         sizeof(attr_info.FileSystemAttributes))))
1290         {
1291                 vol_flags = attr_info.FileSystemAttributes;
1292         } else {
1293                 winnt_warning(status, L"\"%ls\": Can't get volume attributes",
1294                               printable_path(full_path));
1295                 vol_flags = 0;
1296         }
1297
1298         /* Get volume ID.  */
1299         status = (*func_NtQueryVolumeInformationFile)(h, &iosb,
1300                                                       &vol_info,
1301                                                       sizeof(vol_info),
1302                                                       FileFsVolumeInformation);
1303         if (likely((NT_SUCCESS(status) || status == STATUS_BUFFER_OVERFLOW) &&
1304                    (iosb.Information >=
1305                         offsetof(FILE_FS_VOLUME_INFORMATION,
1306                                  VolumeSerialNumber) +
1307                         sizeof(vol_info.VolumeSerialNumber))))
1308         {
1309                 params->capture_root_dev = vol_info.VolumeSerialNumber;
1310         } else {
1311                 winnt_warning(status, L"\"%ls\": Can't get volume ID",
1312                               printable_path(full_path));
1313                 params->capture_root_dev = 0;
1314         }
1315         return vol_flags;
1316 }
1317
1318 struct file_info {
1319         u32 attributes;
1320         u32 num_links;
1321         u64 creation_time;
1322         u64 last_write_time;
1323         u64 last_access_time;
1324         u64 ino;
1325         u64 end_of_file;
1326 };
1327
1328 static noinline_for_stack NTSTATUS
1329 get_file_info(HANDLE h, struct file_info *info)
1330 {
1331         IO_STATUS_BLOCK iosb;
1332         NTSTATUS status;
1333         FILE_ALL_INFORMATION all_info;
1334
1335         status = (*func_NtQueryInformationFile)(h, &iosb, &all_info,
1336                                                 sizeof(all_info),
1337                                                 FileAllInformation);
1338
1339         if (unlikely(!NT_SUCCESS(status) && status != STATUS_BUFFER_OVERFLOW))
1340                 return status;
1341
1342         info->attributes = all_info.BasicInformation.FileAttributes;
1343         info->num_links = all_info.StandardInformation.NumberOfLinks;
1344         info->creation_time = all_info.BasicInformation.CreationTime.QuadPart;
1345         info->last_write_time = all_info.BasicInformation.LastWriteTime.QuadPart;
1346         info->last_access_time = all_info.BasicInformation.LastAccessTime.QuadPart;
1347         info->ino = all_info.InternalInformation.IndexNumber.QuadPart;
1348         info->end_of_file = all_info.StandardInformation.EndOfFile.QuadPart;
1349         return STATUS_SUCCESS;
1350 }
1351
1352 static int
1353 winnt_build_dentry_tree_recursive(struct wim_dentry **root_ret,
1354                                   HANDLE cur_dir,
1355                                   wchar_t *full_path,
1356                                   size_t full_path_nchars,
1357                                   const wchar_t *filename,
1358                                   size_t filename_nchars,
1359                                   struct capture_params *params,
1360                                   struct winnt_scan_ctx *ctx)
1361 {
1362         struct wim_dentry *root = NULL;
1363         struct wim_inode *inode = NULL;
1364         HANDLE h = NULL;
1365         int ret;
1366         NTSTATUS status;
1367         struct file_info file_info;
1368         ACCESS_MASK requestedPerms;
1369         u64 sort_key;
1370
1371         ret = try_exclude(full_path, params);
1372         if (unlikely(ret < 0)) /* Excluded? */
1373                 goto out_progress;
1374         if (unlikely(ret > 0)) /* Error? */
1375                 goto out;
1376
1377         /* Open the file.  */
1378         requestedPerms = FILE_READ_DATA |
1379                          FILE_READ_ATTRIBUTES |
1380                          READ_CONTROL |
1381                          ACCESS_SYSTEM_SECURITY |
1382                          SYNCHRONIZE;
1383 retry_open:
1384         status = winnt_openat(cur_dir,
1385                               (cur_dir ? filename : full_path),
1386                               (cur_dir ? filename_nchars : full_path_nchars),
1387                               requestedPerms,
1388                               &h);
1389         if (unlikely(!NT_SUCCESS(status))) {
1390                 if (status == STATUS_DELETE_PENDING) {
1391                         WARNING("\"%ls\": Deletion pending; skipping file",
1392                                 printable_path(full_path));
1393                         ret = 0;
1394                         goto out;
1395                 }
1396                 if (status == STATUS_ACCESS_DENIED &&
1397                     (requestedPerms & FILE_READ_DATA)) {
1398                         /* This happens on encrypted files.  */
1399                         requestedPerms &= ~FILE_READ_DATA;
1400                         goto retry_open;
1401                 }
1402
1403                 winnt_error(status, L"\"%ls\": Can't open file",
1404                             printable_path(full_path));
1405                 if (status == STATUS_FVE_LOCKED_VOLUME)
1406                         ret = WIMLIB_ERR_FVE_LOCKED_VOLUME;
1407                 else
1408                         ret = WIMLIB_ERR_OPEN;
1409                 goto out;
1410         }
1411
1412         /* Get information about the file.  */
1413         status = get_file_info(h, &file_info);
1414         if (!NT_SUCCESS(status)) {
1415                 winnt_error(status, L"\"%ls\": Can't get file information",
1416                             printable_path(full_path));
1417                 ret = WIMLIB_ERR_STAT;
1418                 goto out;
1419         }
1420
1421         if (unlikely(!(requestedPerms & FILE_READ_DATA)) &&
1422             !(file_info.attributes & FILE_ATTRIBUTE_ENCRYPTED))
1423         {
1424                 ERROR("\"%ls\": Permission to read data was denied",
1425                       printable_path(full_path));
1426                 ret = WIMLIB_ERR_OPEN;
1427                 goto out;
1428         }
1429
1430         if (unlikely(!cur_dir)) {
1431                 /* Root of tree being captured; get volume information.  */
1432                 ctx->vol_flags = get_volume_information(h, full_path, params);
1433                 params->capture_root_ino = file_info.ino;
1434         }
1435
1436
1437         /* Create a WIM dentry with an associated inode, which may be shared.
1438          *
1439          * However, we need to explicitly check for directories and files with
1440          * only 1 link and refuse to hard link them.  This is because Windows
1441          * has a bug where it can return duplicate File IDs for files and
1442          * directories on the FAT filesystem.
1443          *
1444          * Since we don't follow mount points on Windows, we don't need to query
1445          * the volume ID per-file.  Just once, for the root, is enough.  But we
1446          * can't simply pass 0, because then there could be inode collisions
1447          * among multiple calls to win32_build_dentry_tree() that are scanning
1448          * files on different volumes.  */
1449         ret = inode_table_new_dentry(params->inode_table,
1450                                      filename,
1451                                      file_info.ino,
1452                                      params->capture_root_dev,
1453                                      (file_info.num_links <= 1),
1454                                      &root);
1455         if (ret)
1456                 goto out;
1457
1458         /* Get the short (DOS) name of the file.  */
1459         status = winnt_get_short_name(h, root);
1460
1461         /* If we can't read the short filename for any reason other than
1462          * out-of-memory, just ignore the error and assume the file has no short
1463          * name.  This shouldn't be an issue, since the short names are
1464          * essentially obsolete anyway.  */
1465         if (unlikely(status == STATUS_NO_MEMORY)) {
1466                 ret = WIMLIB_ERR_NOMEM;
1467                 goto out;
1468         }
1469
1470         inode = root->d_inode;
1471
1472         if (inode->i_nlink > 1) {
1473                 /* Shared inode (hard link); skip reading per-inode information.
1474                  */
1475                 goto out_progress;
1476         }
1477
1478         inode->i_attributes = file_info.attributes;
1479         inode->i_creation_time = file_info.creation_time;
1480         inode->i_last_write_time = file_info.last_write_time;
1481         inode->i_last_access_time = file_info.last_access_time;
1482
1483         /* Get the file's security descriptor, unless we are capturing in
1484          * NO_ACLS mode or the volume does not support security descriptors.  */
1485         if (!(params->add_flags & WIMLIB_ADD_FLAG_NO_ACLS)
1486             && (ctx->vol_flags & FILE_PERSISTENT_ACLS))
1487         {
1488                 status = winnt_get_security_descriptor(h, inode,
1489                                                        params->sd_set, ctx,
1490                                                        params->add_flags);
1491                 if (!NT_SUCCESS(status)) {
1492                         winnt_error(status,
1493                                     L"\"%ls\": Can't read security descriptor",
1494                                     printable_path(full_path));
1495                         ret = WIMLIB_ERR_STAT;
1496                         goto out;
1497                 }
1498         }
1499
1500         /* If this is a reparse point, load the reparse data.  */
1501         if (unlikely(inode->i_attributes & FILE_ATTRIBUTE_REPARSE_POINT)) {
1502                 ret = winnt_load_reparse_data(h, inode, full_path, params);
1503                 if (ret)
1504                         goto out;
1505         }
1506
1507         sort_key = get_sort_key(h);
1508
1509         if (unlikely(inode->i_attributes & FILE_ATTRIBUTE_ENCRYPTED)) {
1510                 /* Load information about the raw encrypted data.  This is
1511                  * needed for any directory or non-directory that has
1512                  * FILE_ATTRIBUTE_ENCRYPTED set.
1513                  *
1514                  * Note: since OpenEncryptedFileRaw() fails with
1515                  * ERROR_SHARING_VIOLATION if there are any open handles to the
1516                  * file, we have to close the file and re-open it later if
1517                  * needed.  */
1518                 (*func_NtClose)(h);
1519                 h = NULL;
1520                 ret = winnt_scan_efsrpc_raw_data(inode, full_path,
1521                                                  params->unhashed_blobs);
1522                 if (ret)
1523                         goto out;
1524         } else {
1525                 /*
1526                  * Load information about data streams (unnamed and named).
1527                  *
1528                  * Skip this step for encrypted files, since the data from
1529                  * ReadEncryptedFileRaw() already contains all data streams (and
1530                  * they do in fact all get restored by WriteEncryptedFileRaw().)
1531                  *
1532                  * Note: WIMGAPI (as of Windows 8.1) gets wrong and stores both
1533                  * the EFSRPC data and the named data stream(s)...!
1534                  */
1535                 ret = winnt_scan_data_streams(h,
1536                                               full_path,
1537                                               full_path_nchars,
1538                                               inode,
1539                                               params->unhashed_blobs,
1540                                               file_info.end_of_file,
1541                                               ctx->vol_flags);
1542                 if (ret)
1543                         goto out;
1544         }
1545
1546         if (unlikely(should_try_to_use_wimboot_hash(inode, ctx, params))) {
1547                 ret = try_to_use_wimboot_hash(h, inode, params->blob_table, ctx,
1548                                               full_path);
1549                 if (ret)
1550                         goto out;
1551         }
1552
1553         set_sort_key(inode, sort_key);
1554
1555         if (inode_is_directory(inode)) {
1556
1557                 /* Directory: recurse to children.  */
1558
1559                 if (unlikely(!h)) {
1560                         /* Re-open handle that was closed to read raw encrypted
1561                          * data.  */
1562                         status = winnt_openat(cur_dir,
1563                                               (cur_dir ?
1564                                                filename : full_path),
1565                                               (cur_dir ?
1566                                                filename_nchars : full_path_nchars),
1567                                               FILE_LIST_DIRECTORY | SYNCHRONIZE,
1568                                               &h);
1569                         if (!NT_SUCCESS(status)) {
1570                                 winnt_error(status,
1571                                             L"\"%ls\": Can't re-open file",
1572                                             printable_path(full_path));
1573                                 ret = WIMLIB_ERR_OPEN;
1574                                 goto out;
1575                         }
1576                 }
1577                 ret = winnt_recurse_directory(h,
1578                                               full_path,
1579                                               full_path_nchars,
1580                                               root,
1581                                               params,
1582                                               ctx);
1583                 if (ret)
1584                         goto out;
1585         }
1586
1587 out_progress:
1588         params->progress.scan.cur_path = printable_path(full_path);
1589         if (likely(root))
1590                 ret = do_capture_progress(params, WIMLIB_SCAN_DENTRY_OK, inode);
1591         else
1592                 ret = do_capture_progress(params, WIMLIB_SCAN_DENTRY_EXCLUDED, NULL);
1593 out:
1594         if (likely(h))
1595                 (*func_NtClose)(h);
1596         if (unlikely(ret)) {
1597                 free_dentry_tree(root, params->blob_table);
1598                 root = NULL;
1599                 ret = report_capture_error(params, ret, full_path);
1600         }
1601         *root_ret = root;
1602         return ret;
1603 }
1604
1605 static void
1606 winnt_do_scan_warnings(const wchar_t *path, const struct winnt_scan_ctx *ctx)
1607 {
1608         if (likely(ctx->num_get_sacl_priv_notheld == 0 &&
1609                    ctx->num_get_sd_access_denied == 0))
1610                 return;
1611
1612         WARNING("Scan of \"%ls\" complete, but with one or more warnings:", path);
1613         if (ctx->num_get_sacl_priv_notheld != 0) {
1614                 WARNING("- Could not capture SACL (System Access Control List)\n"
1615                         "            on %lu files or directories.",
1616                         ctx->num_get_sacl_priv_notheld);
1617         }
1618         if (ctx->num_get_sd_access_denied != 0) {
1619                 WARNING("- Could not capture security descriptor at all\n"
1620                         "            on %lu files or directories.",
1621                         ctx->num_get_sd_access_denied);
1622         }
1623         WARNING("To fully capture all security descriptors, run the program\n"
1624                 "          with Administrator rights.");
1625 }
1626
1627 #define WINDOWS_NT_MAX_PATH 32768
1628
1629 /* Win32 version of capturing a directory tree.  */
1630 int
1631 win32_build_dentry_tree(struct wim_dentry **root_ret,
1632                         const wchar_t *root_disk_path,
1633                         struct capture_params *params)
1634 {
1635         wchar_t *path;
1636         int ret;
1637         UNICODE_STRING ntpath;
1638         struct winnt_scan_ctx ctx;
1639         size_t ntpath_nchars;
1640
1641         /* WARNING: There is no check for overflow later when this buffer is
1642          * being used!  But it's as long as the maximum path length understood
1643          * by Windows NT (which is NOT the same as MAX_PATH).  */
1644         path = MALLOC((WINDOWS_NT_MAX_PATH + 1) * sizeof(wchar_t));
1645         if (!path)
1646                 return WIMLIB_ERR_NOMEM;
1647
1648         ret = win32_path_to_nt_path(root_disk_path, &ntpath);
1649         if (ret)
1650                 goto out_free_path;
1651
1652         if (ntpath.Length < 4 * sizeof(wchar_t) ||
1653             ntpath.Length > WINDOWS_NT_MAX_PATH * sizeof(wchar_t) ||
1654             wmemcmp(ntpath.Buffer, L"\\??\\", 4))
1655         {
1656                 ERROR("\"%ls\": unrecognized path format", root_disk_path);
1657                 ret = WIMLIB_ERR_INVALID_PARAM;
1658         } else {
1659                 ntpath_nchars = ntpath.Length / sizeof(wchar_t);
1660                 wmemcpy(path, ntpath.Buffer, ntpath_nchars);
1661                 path[ntpath_nchars] = L'\0';
1662
1663                 params->capture_root_nchars = ntpath_nchars;
1664                 if (path[ntpath_nchars - 1] == L'\\')
1665                         params->capture_root_nchars--;
1666                 ret = 0;
1667         }
1668         HeapFree(GetProcessHeap(), 0, ntpath.Buffer);
1669         if (ret)
1670                 goto out_free_path;
1671
1672         memset(&ctx, 0, sizeof(ctx));
1673
1674         ret = winnt_build_dentry_tree_recursive(root_ret, NULL,
1675                                                 path, ntpath_nchars,
1676                                                 L"", 0, params, &ctx);
1677 out_free_path:
1678         FREE(path);
1679         if (ret == 0)
1680                 winnt_do_scan_warnings(root_disk_path, &ctx);
1681         return ret;
1682 }
1683
1684 #endif /* __WIN32__ */